City Research Online

Abstract

As machine learning systems become increasingly integrated into critical decision-making processes, the need for clear and comprehensible model explanations has grown significantly. Traditional interpretability methods, such as rule-based models and feature attribution techniques, each offer unique strengths but also face significant limitations when used independently. Rule-based approaches provide intuitive, logic-driven explanations but can be difficult to scale. In contrast, feature attribution methods like SHAP can work well with large amounts of data and complex models, yet often lack clarity and alignment with human reasoning. In this paper, we propose a hybrid human-centric approach that combines rule-based and feature attribution techniques in a visual way to deliver more coherent and actionable explanations. By extracting and analyzing rules from a Random Forest classifier and integrating SHAP values to guide rule exploration, we offer a unified framework that enhances interpretability. We demonstrate the effectiveness of this approach through a case study on a vessel trajectory classification task, highlighting how the combination of rule distributions and feature attributions can provide deeper insight into model behavior and decision-making processes.

Publication Type:	Conference or Workshop Item (Paper)
Additional Information:	Copyright © 2025 authors. This paper is published under the Creative Commons License Attribution 4.0 International (CC BY 4.0).
Publisher Keywords:	Explainable AI, Visual Analytics, Rule-based, Attribute-based
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science T Technology > TA Engineering (General). Civil engineering (General)
Departments:	School of Science & Technology School of Science & Technology > Department of Computer Science
SWORD Depositor:	Symplectic Administrator

Export

Downloads